I know some of you are data nerds and suckers for statistics, as am I. Feel free to use this thread liberally to post data and crunch numbers.
Nobody will confirm this, but I believe that Baltimore City doesn't complete its building permit survey for reports properly (or perhaps at all.) The number of building permits per 1,000 was probably around 4.I have two questions about census data that perhaps someone can answer for me.
1. Can someone elaborate on some of the differences between the Decennial Census and the American Community Survey? Often, the numbers presented in each are so materially different that it is difficult to compare them.
As an example, consider Baltimore’s downtown census tract – 40100. The Decennial Census gave it a 2010 population of 4,006 people. The ACS gave it a 2010 population of 2,996. According to the ACS, population in that census tract has grown every year, and had a 2016 population of 3,868 people.
The 2010 ACS population was just 75 percent of the 2010 Decennial population. The ACS population grew 29 percent from 2010-2016, but its 2016 population was still only 96 percent of the 2010 Decennial population. If you were to compare ACS numbers with ACS numbers, you would see a population increase, but if you compare ACS with Dicennial, you see a decrease.
Is one more correct than the other?
2. I saw this graph posted on social media. It shows that Baltimore had the lowest number of residential permits issued per capita in 2017 of the major cities listed. Supposedly, the data comes from the Census Building Permits Survey. I was unable to recreate the data, but the chart seems to suggest that Baltimore issued just 600 or fewer housing units last year. That seemed unlikely to me. Anyone have any insight on that?
I looked it up, and as best I can figure out, the city reported permits for 438 housing units in 2017. Gotta be right... its the Census!Nobody will confirm this, but I believe that Baltimore City doesn't complete its building permit survey for reports properly (or perhaps at all.) The number of building permits per 1,000 was probably around 4.
In response to question 1: I treat the ACS and decennial Census number of similar but different counts that will never ever match up. I prefer to cobble post 2010 ACS trends onto 2010 counts for what I think are the most reliable numbers.
Not plausible at all. Its hard to say exactly when I should count a count a construction start, but my best guess estimate is 2,250 to 2,500 in 2017 and almost that many the year before. I believe that this year will be below 2,000.Thanks for shedding some light on those questions.
Am I correct that you track permits as well? Do those numbers sound plausible to you?
Do these permit numbers factor in any way into the ACS population estimates? If the city is not submitting this information correctly or at all, and the counts are lower than they should be, would have affect the population estimates in any way?Not plausible at all. Its hard to say exactly when I should count a count a construction start, but my best guess estimate is 2,250 to 2,500 in 2017 and almost that many the year before. I believe that this year will be below 2,000.
The study showed that the city has gained 23,228 college educated residents over age 25 since 2010, which reflects a nearly a 23 percent increase over the prior decade.
Are more vacant units being demolished than are being completed? How many of the 52,000 unit difference are places that can actually be lived in?The Census Bureau’s Housing Unit estimates were released in May. For Baltimore, it shows a decline of 294 units over the past year.
Here are the numbers going back to 2010:
2010 296,685 (Decennial Census)
2010 296,712 (Estimate Base)
This compares interestingly to an increase of a little more than 4,000 households between 2010 and 2016.
Certainly, these numbers are compatible; there are still 50,000 more housing units than there are households. I have no idea if that is too many, too few, or just right. It makes sense that inventory should exceed occupancy to some degree. I compared it to notoriously-housing-crunched San Francisco County, which had 356,797 households in 2016 and 397,550 housing units, or a ratio of 0.89 compared to 0.82 for Baltimore.
In any event, I am more interested in Baltimore’s decline in housing units, a problem that SF doesn’t have, of course.
Losing 294 units in 2016 struck me as odd, since we seem to be building several hundreds to several thousand new units every year. I can’t imagine we’re losing that many more. Even if we didn’t build a single new unit, that loss equates to losing 1 out of every 1,000 units. That seems really high, and is quite a bit higher than the 2017 Housing Unit Loss Rates for the Northeast Region that the Census Bureau uses to calculate its estimates.
I wonder if this is another instance that might stem from Baltimore not reporting its new housing numbers correctly.